Configure your appliance

Size the appliance you will own.

A turnkey AI server that lives inside your building. Answer three questions for a starting point, or build it yourself. Every option stays valid, and you see the spec, the fit and an indicative price as you go.

01How many people will use it?
02What is the main workload?
03Where will it live?
Or start from a tier

Pick the shape that fits where your AI will live. This sets the GPU range, cooling and power downstream.

Choose the GPU tier and how many. The configurator resolves the correct card edition automatically so every build is valid.

GPU tier
Quantity
2
Range for Tower with PRO 4500: 1–2
2x PRO 4500 air-cooled Workstation edition.
Aggregate VRAM across 2 cards64 GB

Cooling is resolved from the form factor and GPU tier, so every build is valid.

Air. The PRO 4500 is air-cooled and silent.

Choose the processor. We mark the class we would recommend for your GPU count; you can pick any in range. The exact AMD EPYC part shows in the spec sheet.

Dual-socket processors are available on the rack-mount server.

DDR5-5600 RDIMM. System RAM is for headroom and data handling, not inference speed; the workload is VRAM-bandwidth-bound.

NVMe storage with a mirrored boot pair as standard. The data pool sets redundancy for your models and indexes.

Capacity (raw)
Data pool redundancy
Usable capacity: 2 TB from 4 TB raw.

Network interface for the appliance. Higher tiers suit multi-appliance scale-out.

Power and placement. The appliance ships worldwide; set the mains and connector for where it will live.

Single PSU. Sized for the console and its liquid loop.
Region & voltage
  • Continental Europe230 V · Schuko / CEE · IEC C13–C19
  • United Kingdom & Ireland230 V · BS 1363 · IEC C13–C19
  • North America110–120 V / 208 V · NEMA 5-15 / L6-30
  • Latin America110–220 V · NEMA / IEC C13
  • Japan100 V / 200 V · JIS C 8303 / NEMA L6
  • Asia-Pacific220–240 V · AS/NZS 3112 · IEC C13–C19
  • Middle East & Africa220–240 V · IEC C13–C19 · PDU

Half the value of the appliance. Everything here ships pre-installed and tuned.

Malogica Mind
Private knowledge assistant over your own documents.
Malogica Agents
Visual, no-code automation that runs on your hardware.
Curated open models
A pre-loaded, tuned set of open-weight models.
Pre-loaded model set
Up to ~70B for a strong knowledge assistant. Needs ~80 GB resident · build has 64 GB.
The medium set needs about 80 GB but this build has 64 GB. Choose a smaller set, add GPUs, or step up to higher-VRAM cards.

Support tier and warranty. Remote assistance only ever runs over a secure channel you control.

Support SLA
Warranty